torch

torch.nn.module?? Full code
torch.nn.module? shift+tab (about)
%%bash
v @ w (scalar pdt)
x = torch.Tensor(5).random_(10);
_ inplace operator
dim(), numel(), 

------Math-----
convolutional, orthogonal

autograd

how to check that the params are relevant?
-If the loss function is differentiable based on theta (gradient descent, hunt for the global minima)

--12th feb--
y we use y combinator in backprop?
pooling useful with slight changes in an img./ speech locally invariant/ small lipschid constants
---13th feb--
fcl to have global statistics
1d convo conversion
___20th feb---
Mini batch - adding some sort of noise
Batch norm over dropouts

---5th march---
IPAM new methods in dl